chore: support loglevel when start server (#1636)

* chore: support loglevel when start server

* Fix comment and set logger for engine

* Update docs

Author: nguyenhoangthuan99

docs/docs/cli/start.mdx

@@ -34,6 +34,8 @@ You can use the `--verbose` flag to display more detailed output of the internal
 | Option                       | Description                               | Required | Default value | Example                       |
 | ---------------------------- | ----------------------------------------- | -------- | ------------- | ----------------------------- |
 | `-h`, `--help`               | Display help information for the command.  | No       | -             | `-h`                          |
-| `-p`, `--port <port>`        | Port to serve the application.             | No       | `39281`       | `-p 39281`                     |
+| `-p`, `--port <port>`        | Port to serve the application.             | No       | -             | `-p 39281`                     |
+| `--loglevel <loglevel>`        | Setup loglevel for cortex server, supported levels are TRACE, DEBUG, INFO, WARN, ERROR             | No       | -             | `--loglevel DEBUG`                     |
+
 <!-- | `-a`, `--address <address>`  | Address to use.                            | No       | -             | `-a 192.168.1.1`              | -->
 <!--| `--dataFolder <dataFolder>`  | Set the data folder directory             | No       | -             | `--dataFolder /path/to/data`  | -->

engine/cli/command_line_parser.cc

@@ -468,6 +468,16 @@ void CommandLineParser::SetupSystemCommands() {
   start_cmd->group(kSystemGroup);
   cml_data_.port = std::stoi(cml_data_.config.apiServerPort);
   start_cmd->add_option("-p, --port", cml_data_.port, "Server port to listen");
+  start_cmd->add_option("--loglevel", cml_data_.log_level,
+                        "Set up log level for server, accepted TRACE, DEBUG, "
+                        "INFO, WARN, ERROR");
+  if (cml_data_.log_level != "INFO" && cml_data_.log_level != "TRACE" &&
+      cml_data_.log_level != "DEBUG" && cml_data_.log_level != "WARN" &&
+      cml_data_.log_level != "ERROR") {
+    CLI_LOG("Invalid log level: " << cml_data_.log_level
+                                  << ", Set Loglevel to INFO");
+    cml_data_.log_level = "INFO";
+  }
   start_cmd->callback([this] {
     if (std::exchange(executed_, true))
       return;
@@ -484,7 +494,7 @@ void CommandLineParser::SetupSystemCommands() {
     }
     commands::ServerStartCmd ssc;
     ssc.Exec(cml_data_.config.apiServerHost,
-             std::stoi(cml_data_.config.apiServerPort));
+             std::stoi(cml_data_.config.apiServerPort), cml_data_.log_level);
   });
 
   auto stop_cmd = app_.add_subcommand("stop", "Stop the API server");

engine/cli/command_line_parser.h

@@ -58,8 +58,11 @@ class CommandLineParser {
     bool display_engine = false;
     bool display_version = false;
     std::string filter = "";
+    std::string log_level = "INFO";
+
     bool show_menu = false;
 
+
     int port;
     config_yaml_utils::CortexConfig config;
     std::unordered_map<std::string, std::string> model_update_options;

engine/cli/commands/server_start_cmd.cc

@@ -26,7 +26,14 @@ bool TryConnectToServer(const std::string& host, int port) {
 
 ServerStartCmd::ServerStartCmd() {}
 
-bool ServerStartCmd::Exec(const std::string& host, int port) {
+bool ServerStartCmd::Exec(const std::string& host, int port,
+                          const std::optional<std::string>& log_level) {
+  std::string log_level_;
+  if (!log_level.has_value()) {
+    log_level_ = "INFO";
+  } else {
+    log_level_ = log_level.value();
+  }
   auto exe = commands::GetCortexServerBinary();
   auto get_config_file_path = []() -> std::string {
     if (file_manager_utils::cortex_config_file_path.empty()) {
@@ -53,6 +60,7 @@ bool ServerStartCmd::Exec(const std::string& host, int port) {
   std::string params = "--start-server";
   params += " --config_file_path " + get_config_file_path();
   params += " --data_folder_path " + get_data_folder_path();
+  params += " --loglevel " + log_level_;
   std::string cmds = cortex_utils::GetCurrentPath() + "/" + exe + " " + params;
   // Create child process
   if (!CreateProcess(
@@ -107,7 +115,7 @@ bool ServerStartCmd::Exec(const std::string& host, int port) {
     std::string p = cortex_utils::GetCurrentPath() + "/" + exe;
     execl(p.c_str(), exe.c_str(), "--start-server", "--config_file_path",
           get_config_file_path().c_str(), "--data_folder_path",
-          get_data_folder_path().c_str(), (char*)0);
+          get_data_folder_path().c_str(), "--loglevel", log_level_.c_str(), (char*)0);
   } else {
     // Parent process
     if (!TryConnectToServer(host, port)) {

engine/cli/commands/server_start_cmd.h

@@ -2,7 +2,7 @@
 
 #include <string>
 #include "httplib.h"
-
+#include <optional>
 namespace commands {
 
 inline bool IsServerAlive(const std::string& host, int port) {
@@ -17,6 +17,6 @@ inline bool IsServerAlive(const std::string& host, int port) {
 class ServerStartCmd {
  public:
   ServerStartCmd();
-  bool Exec(const std::string& host, int port);
+  bool Exec(const std::string& host, int port, const std::optional<std::string>& log_level = std::nullopt);
 };
 }  // namespace commands

engine/controllers/server.cc

@@ -27,6 +27,7 @@ void server::ChatCompletion(
   LOG_DEBUG << "Start chat completion";
   auto json_body = req->getJsonObject();
   bool is_stream = (*json_body).get("stream", false).asBool();
+  LOG_DEBUG << "request body: " << json_body->toStyledString();
   auto q = std::make_shared<services::SyncQueue>();
   auto ir = inference_svc_->HandleChatCompletion(q, json_body);
   if (ir.has_error()) {
@@ -141,7 +142,7 @@ void server::ProcessStreamRes(std::function<void(const HttpResponsePtr&)> cb,
     }
 
     auto str = res["data"].asString();
-    LOG_TRACE << "data: " << str;
+    LOG_DEBUG << "data: " << str;
     std::size_t n = std::min(str.size(), buf_size);
     memcpy(buf, str.data(), n);
 
@@ -157,6 +158,7 @@ void server::ProcessNonStreamRes(std::function<void(const HttpResponsePtr&)> cb,
                                  services::SyncQueue& q) {
   auto [status, res] = q.wait_and_pop();
   function_calling_utils::PostProcessResponse(res);
+  LOG_DEBUG << "response: " << res.toStyledString();
   auto resp = cortex_utils::CreateCortexHttpJsonResponse(res);
   resp->setStatusCode(
       static_cast<drogon::HttpStatusCode>(status["status_code"].asInt()));

engine/cortex-common/EngineI.h

@@ -4,7 +4,7 @@
 #include <memory>
 
 #include "json/value.h"
-
+#include "trantor/utils/Logger.h"
 class EngineI {
  public:
   virtual ~EngineI() {}
@@ -36,4 +36,5 @@ class EngineI {
 
   virtual bool SetFileLogger(int max_log_lines,
                              const std::string& log_path) = 0;
+  virtual void SetLogLevel(trantor::Logger::LogLevel logLevel) = 0;
 };

engine/main.cc

@@ -57,7 +57,6 @@ void RunServer(std::optional<int> port) {
   }
   std::cout << "Host: " << config.apiServerHost
             << " Port: " << config.apiServerPort << "\n";
-
   // Create logs/ folder and setup log to file
   std::filesystem::create_directories(
       std::filesystem::path(config.logFolderPath) /
@@ -181,6 +180,9 @@ int main(int argc, char* argv[]) {
       file_manager_utils::cortex_data_folder_path = argv[i + 1];
     } else if (strcmp(argv[i], "--port") == 0) {
       server_port = std::stoi(argv[i + 1]);
+    } else if (strcmp(argv[i], "--loglevel") == 0) {
+      std::string log_level = argv[i + 1];
+      logging_utils_helper::SetLogLevel(log_level);
     }
   }
 

engine/services/engine_service.cc

@@ -832,6 +832,11 @@ cpp::result<void, std::string> EngineService::LoadEngine(
     } else {
       CTL_WRN("Method SetFileLogger is not supported yet");
     }
+    if (en->IsSupported("SetLogLevel")) {
+      en->SetLogLevel(trantor::Logger::logLevel());
+    } else {
+      CTL_WRN("Method SetLogLevel is not supported yet");
+    }
   }
   CTL_DBG("Loaded engine: " << ne);
   return {};

engine/services/inference_service.cc

@@ -240,4 +240,4 @@ bool InferenceService::HasFieldInReq(std::shared_ptr<Json::Value> json_body,
   }
   return true;
 }
-}  // namespace services
+}  // namespace services